Leveraging U-Net and Convnext for Accurate Plant Disease Prediction with Hyperparameter Tuning
Bibliographic record
Abstract
Early and immediate detection of plant diseases is essential for crop protection and food security. Usually, these may be delayed and adversely impact agriculture because inspections are subjective and time-consuming. This research proposes a deep learning-based framework consisting of the combination of U-Net with Attention Gates for an accurate image segmentation process that hands over the task of disease classification to ConvNeXt. Image preprocessing using CLAHE and augmentations are introduced to tackle some dataset problems, such as noisy data and variations within plant species. Hyperparameter optimization using Optuna is set in place for fine-tuning along with learning rate and batch size, which results in even further improvement in performance. The performance of the proposed model is found better than those of baseline models like VGG-19, MobileNetV2, and EfficientNet with accuracies of 99.16%, 98.48% precision, 98.22% recall, and 98.33% F1-score. Despite its high success rate, the analysis exposes the limitations of the proposed system through ROC curve and confusion matrix analysis, indicating further avenues for improvements. Even then are good levels of robustness and generalization present in the model across different plant species and diseases, thus offering a good discriminatory power for disease diagnosis in actual-farm environments. This study paves the way for developing smart, fully automated systems for early disease detection, which can play a significant role in enhancing global food security and improving crop yields.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.002 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".